iv THE CHURCH OF S. IRENE 103
parts representing different periods solves also the problem of the elliptical domical vault. For it is difficult to imagine that a Byzantine architect with a free hand would choose to build such a vault. But given the supports Mr. George believes were left standing after the earthquake of 740, and given also the narthex on the west, the architect's liberty was limited, and he would be forced to cover the space thus bounded in the best way the circumstances allowed.
How the western portion of the church was roofed in Justinian's time it is impossible to say with certainty. There are buttress slips in the south wall at gallery level and in the nave below, where the break occurs in the arcade, that suggest the existence, in the church as originally built by Justinian, of a narthex carrying a gallery. In that case the length of the barrel vault over the western part of the church would be about the length of the barrel vault over the eastern part5 and the church would then show in plan a regular cross with a dome at the centre, two lateral doors, one of which is now built up, giving access to the ends of the narthex,
The dates here assigned to the different parts of the building simplify the problem of the tall drum below the main dome. That this could have been built by Justinian, as has been supposed, is difficult of belief if the large domes which are known to have been built by him are carefully examined. It is true that the drum dome of S. Sophia, Salonica, has also been claimed for Justinian, but that drum is low and only partially developed, and although its date is not known, the consensus of opinion is against its being so early. The whole question of the development of the drum still awaits treatment at the hands of an investigator who has thoroughly studied the buildings themselves, and perhaps the publication of the results obtained by Mr. George at S. Sophia, Salonica, and S. Irene, Constantinople, two crucial examples, will throw some light on the subject. For the present the date here given for the drum of S. Irene (i.e. towards the middle of the eighth century) is an inherently probable one,
In the foregoing description of S. Irene there is no